PASO CUATRO: CREE UN ESBOZO DEL CAPÍTULO

The dependent variable is a count of ratification incidence of seven of the UN core human rights agreements with a Poisson distribution. A series of the goodness-of-fit tests indicate the

presence of over-dispersion in the empirical models in this chapter. Thus, a negative binomial regression model, which allows a conditional variance to exceed the conditional means with a Poisson distribution, is used to estimate the results. With a panel data set-up, I also use a random-effects negative binomial regression model to address unobservable heterogeneity

among countries. Most covariates are lagged one year except for time invariant variables. I cluster robust standard errors on the autocratic countries to address the possible non-

independence of observations within the same country. For robustness checks, I estimated the same model with a logit and random effects logit model to see if the results are driven by the binary nature of ratifications instead of that of count.

Figure 3.1 provides suggestive evidence from descriptive statistics to support my

theoretical expectations. The left plot shows that around 13 percent and 18 percent of autocratic leaders, who inherited support bases from outgoing leaders, ratify one and more than one UN core human rights treaties, respectively, while in power. More than a majority of such autocrats (e.g. around 69 percent) do not ratify any human rights treaties. Compared to this, the right plot displays slightly different patterns. Autocrats who transitioned with new leader support coalition ratify at least one such treaty at 10 percentage points greater than those with the same support coalition.

FIGURE 3.1. Frequency of autocratic human rights treaty ratifications by change

in domestic sources of autocratic leader support.

NOTE: The left plot displays the frequency and fraction of multiple ratifications of human rights treaties by an autocrat who enters into power without changes in domestic sources of political and social supports from his predecessor. The right plot examines autocrats who achieve and retain power with different domestic sources of support to stay in power from his predecessor. Data: Svolik (2012a), CHISOLS (Mattes et al.

forthcoming), and the UNTS (2016).

Next, I test the hypotheses with the multivariate models. Table 3.1 reports the primary empirical results regarding H3-1. SOLS change variable is positive and statistically significant at the 1 or 5 percent levels in all model estimations. The effect is also substantively significant, as shown in Figure 3.2. Autocrats with changes in domestic support coalitions are likely to ratify 42.4 percent more human rights treaties than those without such changes in a year. This

difference in the expected count of ratification incidences is statistically significant with the p- value of 0.018. Control variables generally perform as expected and are consistently across different models. When more human rights treaties become available for ratification in a given

68.9 12.79 18.31 59.46 16.67 23.87 0 20 40 60 80

none rat =1 rat > 1 none rat = 1 rat >1

no change change p e rce n ta g e

year, an autocrat is likely to ratify a higher number of human rights treaties, and this is

statistically significant at the 1 percent level across all model estimations. Under the common law legal system, autocrats are less likely to ratify a higher number of human rights treaties, and this effect is statistically significant at the 5 percent level across all models. Other control variables, such as multiparty, independent judiciary, latent human rights scores, Muslim

populations, and regional autocratic ratification rates, never achieve statistical significance from models 1 through 4 in Table 3.1.

TABLE 3.1: Multiple Ratifications of Core UN Human Rights Treaties, 1966-2008. ratification incidence NB

(1) RENB (2) logit (3) RE logit (4) SOLS changet 0.353** 0.461*** 0.409*** 0.494***

(0.142) (0.164) (0.153) (0.191) number of available HRAs 0.363*** 0.422*** 0.296*** 0.474*** for ratificationt (0.050) (0.064) (0.051) (0.086)

multipartyt-1 0.217 0.237 0.154 0.219

(0.178) (0.162) (0.190) (0.194) independent judiciaryt-1 0.043 0.024 0.069 0.040

(0.111) (0.115) (0.129) (0.135) latent human rights scoret-1 -0.001 0.035 0.010 0.084

(0.097) (0.095) (0.095) (0.114) common lawt -0.487** -0.541*** -0.466** -0.610** (0.196) (0.201) (0.215) (0.245) muslimt 0.002 0.002 0.003 0.002 (0.002) (0.002) (0.002) (0.003) regional autocratic 0.028 0.050 -0.003 0.027 ratificationst-1 (0.063) (0.064) (0.058) (0.072) constant -3.451*** -2.640*** -3.259*** -3.936*** (0.378) (0.398) (0.365) (0.485) ln α 0.915*** 2.548*** (0.174) (0.417) N 1,959 1,959 1,959 1,959 N of countries 115 115

NOTE: Robust standard errors clustered on country in parentheses except for the random effects models in Models (2) and (4). *** p<0.01, **p<0.05, *p<0.10.

FIGURE 3.2: Change in predicted count of ratification by SOLS

change, 1966-2008. Based on NB model (1) in Table 3.1.

For robustness checks, I added additional control variates that might be associated with multiple ratifications either positively (e.g., trade dependence and amount of received foreign aid) or negatively (e.g., the presence of civil wars). Autocrats who rely on external economic pressures are expected to proactively pursue human rights treaty ratifications (Garriga 2016; Hathaway 2005; Smith Cannoy 2012). Conversely, autocrats who are either fighting or just fought civil conflicts are less likely to seek out human rights treaty

memberships because governments themselves tend to commit human rights violations during civil wars (Nalepa 2010). Table 3A.1 in the Appendix at the end of this chapter reports the empirical results. The impact of SOLS change stays intact and statistically significant across all model estimations. The additional covariates behave interestingly. While the presence of civil war fails to achieve statistical significance, two covariates related to external pressure are positively associated with multiple ratifications of human rights treaties. Autocrats’ vulnerability and susceptibility to external pressures are beyond the

0 .1 .2 .3 .4 .5 Pre d ict e d co u n t

scope of the current chapter; however, this result yields an interesting potential for future research.

TABLE 3.2: New Leader and Multiple Ratifications of Core

UN Human Rights Treaties, 1966-2008.

ratification incidence NB (1) SOLS changet 0.348** (0.154) new leadert -0.068 (0.259) SOLS changet x new leadert 0.038

(0.357) number of available HRAs 0.363***

for ratificationt (0.050)

multipartyt-1 0.218

(0.178) independent judiciaryt-1 0.045

(0.112) latent human rights scoret-1 -0.002

(0.097) common lawt -0.486** (0.196) muslimt 0.002 (0.002) regional autocratic 0.028 ratificationst-1 (0.062) constant -3.440*** (0.387) ln α 0.915*** (0.174) N 1,959

NOTE: Robust standard errors clustered on country in parentheses except for the random effects models in Models (2) and (4). *** p<0.01, **p<0.05, *p<0.10.

In an additional test, I examine when an autocrat with changes in domestic support coalitions from his predecessor pursues multiple human rights treaty ratifications during his tenure. Table 3.2 provides a supporting evidence to H3-1a. The effect of SOLS change on the

number of ratification incidence is positive and significant only when autocrats are not new leaders. The average marginal effect of SOLS change on the number of ratification is 0.06 after the initial three years of an autocrat’s rule, and this is statistically significant at the 5 percent level. On the other hand, the same is not significant at any conventional levels during the initial three years of an autocrat’s reign.

I switch the structure of the main dataset to make a leader as a unit of analysis to test why autocrats ratify multiple human rights treaties during their tenure, not necessarily in a given year. To do so, I collapse all observations by individual autocrat, keeping years in power and number of human rights treaties available for ratification at their maximum values, multiparty, which is a binary covariate, at its median, ordinal or continuous covariates, such as independent judiciary, latent human rights score, Muslim, and regional autocratic ratifications at their means.

Table 3.3 reports the main result. Controlling for the duration in which an autocrat stays in power as well as the maximum number of available seven core UN human rights treaties, I find that an autocrat with domestic support coalitions different from those of his predecessor is likely to ratify a higher number of human rights treaties during his tenure. The coefficient of SOLS change is positive and statistically significant at the 5 percent level. Substantively, as Figure 3.3 displays, autocrats with distinct domestic support coalitions from their predecessors are likely to ratify around 33% more human rights treaties than those with contiguous domestic support coalitions. This difference is statistically significant at the 5 percent level. The other control variables generally behave as Models in Table 3.1. Autocrats with longer tenure have positive impacts on the number of human rights treaty ratifications. Likewise, as more human rights treaties are available when an autocrat enters into power, the autocrat is more likely to

ratify a higher number of them. Common law is again negatively associated with the number of treaty ratifications. These effects are all significant at the conventional level.

TABLE 3.3: Multiple Ratifications of Core UN Human

Rights Treaties, 1966-2008 (during autocratic tenure)

number of ratifications NB (1) SOLS change 0.286** (0.125) years in power 0.061*** (0.006) number of available HRAs 0.204***

for ratification (0.035)

multiparty 0.171

(0.155)

independent judiciary 0.004

(0.121) latent human rights score 0.009

(0.078) common law -0.323** (0.153) muslim 0.001 (0.001) regional autocratic 0.041 ratifications (0.057) constant -1.660*** (0.299) ln α -1.347*** (0.328) N 278

NOTE: Negative binomial results. Robust standard errors in parentheses. *** p<0.01, **p<0.05, *p<0.10.

FIGURE 3.3: Change in predicted count of ratification by SOLS change,

1966-2008. Based on NB model (1) in Table 3.3.